Advertisement IC Markets

The rollover window is wider than the clock says

Compare ECN/Raw spread brokers, analyze execution speeds, report slippage, and evaluate commission structures for high-frequency traders.
Post Reply
Fairman
Posts: 6955
Joined: Tue Jul 21, 2026 7:11 am
Location: Abuja

The rollover window is wider than the clock says

Post by Fairman »

The rollover window is wider than the clock says

Brokers quote rollover at 5 p.m. New York time, and the platform clock shows some variation of it depending on server settings. Everyone knows to avoid the minute itself.

What gets missed is the shoulder. Spreads on liquid pairs often start stretching 10 to 15 minutes before the official time, because liquidity providers start pulling quotes while their systems prepare for the daily cycle. The recovery on the other side is just as slow. Spreads can stay two or three times normal for up to half an hour after rollover on some brokers, drifting down gradually and not snapping back.

The result is a 45-minute stretch where a scalper is paying an inflated cost on every trade, and the platform shows nothing alarming. A spread of 1.6 on EURUSD instead of 0.6 looks harmless in isolation.

Easy way to see it: record the spread every 10 seconds across rollover for a week and chart the median per time slot. The shape of the bump is surprisingly consistent.

Setting a no-trade window based on the measured bump rather than the clock is the more accurate approach.

How wide was the measured window on the broker being used, and did it differ between EURUSD and a cross like GBPJPY?
It’s Fairman :geek:
Recommended broker for automated trading & scalping IC Markets
FXS London Scalper
Posts: 976
Joined: Sun Oct 18, 2026 12:25 pm

Re: The rollover window is wider than the clock says

Post by FXS London Scalper »

Fairman wrote: Spreads on liquid pairs often start stretching 10 to 15 minutes before the official time, because liquidity providers start pulling quotes while their systems prepare for the daily cycle.
Avoiding the rollover minute isn't enough because spreads start widening before it and stay wide after. Blocking new entries from about fifteen minutes before to fifteen minutes after rollover covers most of the wider window.
FXS Official Team | London Session Desk
Fairman
Posts: 6955
Joined: Tue Jul 21, 2026 7:11 am
Location: Abuja

Re: The rollover window is wider than the clock says

Post by Fairman »

A measured no-trade window is more useful than assuming the spread normalizes at a particular clock time, but the timing needs consistent labels. I would record server timestamps alongside the relevant New York rollover time, since broker clocks and daylight-saving changes can shift the apparent hour. Sampling every ten seconds could capture a broad pattern, though it may miss short-lived spikes; logging the actual spread at each entry and exit is also important for the strategy’s realized cost.

For a week of observations, the median by time slot can describe a typical day, while upper percentiles show how bad the window can get. I would separate major pairs from crosses and compare against each instrument’s own normal spread, rather than apply one absolute pip threshold. A rule could then pause entries when the spread or total expected transaction cost exceeds a pre-set share of the setup’s planned target, and resume only when it remains below the threshold for several observations. That avoids treating a single narrow quote as proof conditions have normalized. The exact window is still a broker- and instrument-specific estimate, so it should be rechecked after execution conditions or account settings change.
It’s Fairman :geek:
Amanhasnoname
Posts: 2060
Joined: Fri Oct 30, 2026 9:45 am

Re: The rollover window is wider than the clock says

Post by Amanhasnoname »

I did exactly this measurement on my main broker last winter, logging the spread every few seconds for two weeks around rollover. On EURUSD the widening started about twelve minutes before the hour and the spread wasn't back to normal until roughly twenty minutes after. On GBPJPY the picture was much worse, starting earlier and staying wide for close to forty minutes on several nights.

Before that I had a simple rule of no trades five minutes either side, and looking back at my log, a good share of my small Asian-session losses came from entries in that shoulder. The spread looked acceptable on the screen, but on targets of two or three pips it was eating a third of the move.

Now I block entries on crosses for an hour around rollover and for about half an hour on the majors. I also remeasure whenever the clocks change, because the window shifted on my broker's server time after daylight saving and caught me out once.
Post Reply